There are conventional techniques which disclose presenting a viewer with a 3D image obtained from any given viewpoint by generating and displaying a virtual image obtained at a place where the camera is not actually found at the time of reproduction, using two two-dimensional (2D) images (hereinafter referred to as “image” unless otherwise noted) each obtained by multiple cameras (see Patent Literature 1, or PTL 1). The technique described in PTL 1 is implemented in the steps below.
First, images obtained by multiple cameras are analyzed, and depth information indicating the depth of an object captured on the images is obtained. Next, each of the images are separated into partial images per depth, and the partial images that are separated into have the depth information assigned to and are outputted as image information. Finally, using trilateration, a right-eye image and a left-eye image are generated from the outputted image information based on the partial images, the depth information, and viewpoint positions of the viewer (a right-eye position and a left-eye position). Each of the right-eye image and the left-eye image is presented to a corresponding one of the eyes of the viewer. As a result, the viewer can three-dimensionally view an object from any given viewpoint.
Furthermore, Non Patent Literature 1, or NPL 1, defines the specifications of the Multi Picture Format (MPO file) that is a file format used for holding a 3D image. MPO files are widely used for digital still cameras (DSC) which obtain 3D still pictures. A single MPO file holds, for example, a left-eye image and a right-eye image each captured by a DSC having two imaging devices.